Oracle Real-Time Decisions (Oracle RTD), Release 3.2, introduces the following new features:
Learning On Choice Attributes and Transient Session Attributes
Max Input Cardinality/Buckets Defined At the Choice Attribute Level
RTD 3.2 now has the option to log detailed decision and choice event information in its schema for transformation into a star schema. RTD provides reporting on top of that star schema with OBIEE dashboards.
RTD now uses a database based topology instead of JGroups. This simplifies RTD cluster configuration and improves stability of large RTD deployments.
RTD has a new service called the Query Service. This service is in charge of providing data for Decision Center and Decision Manager reports. This decouples the handling of Decision Center and Decision Manager Model queries from the active Learning Server and allows the Learning Server to focus on its own tasks. Multiple servers in an RTD cluster can have the Query Service, further enhancing scalability of our reporting infrastructure.
Further scalability improvements have been made on top of the improvements made in 3.0.0.1.11 and later PSUs.
This feature allows to learn on session data at the time of the event while saving everything in one learning record at the end of the session. This reduces the number of learning records needed without any functionality loss. This in turn allows greater scalability on the learning server side.
Max Input Cardinality (for string attributes) and Max Input Buckets (for numeric attributes) can now be overridden at the choice attribute level. This lets people tune how much information is needed in the model at the attribute level.
Scoring Rules can now be externalized in a similar fashion as eligibility and filtering rules.
Refer to the RTD 3.2 certification matrix to see the new certifications in this release. See "Oracle Fusion Middleware Supported System Configurations" for Release 3.2, at:
http://www.oracle.com/technetwork/middleware/ias/downloads/fusion-certification-100350.html